摘要
应用了三种方法将某实际双容水箱液位对象直接辨识为双容时滞传递函数模型,并进行了对比分析,探讨了其工程应用性问题。由双容水箱实验装置获取阶跃响应数据,分别应用最小二乘法、粒子群算法和差分进化算法辨识出该类模型的4个参数。辨识结果表明,差分进化算法的辨识精度最高,且算法参数相对于粒子群算法较容易设置,辨识过程耗时最长;粒子群算法的辨识精度次之,算法参数最不易设置;最小二乘法算法简单,容易实现,耗时短,辨识精度最低。
In the paper, three methods were applied to a two-tank level process which was directly identified as two-capacity time delay transfer function, and some engineering application problems were discussed after comparison and analysis. The step response level data were obtained from two-tank experiment appliance, and then Least Square (LS) method, Particle Swarm Optimization (PSO) method and Differential Evolution (DE) method were applied to identify the four parameters of the transfer function. The identification results show that the DE method has very high precision, and the algorithm parameter is relatively easier to set compared with PSO method, but the identification time is the longest; the precision of PSO method is the second, but the algorithm parameters are the most difficult to set; the LS method is easiest to implemented with least time loss and higher identification precision for time delay, but the worst whole identification precision.
出处
《计算机仿真》
CSCD
北大核心
2015年第11期407-410,共4页
Computer Simulation
基金
上海市科学技术委员会工程技术研究中心项目(14DZ2251100)
上海市"科技创新行动计划"重点科研项目(13111104300)